🚀 Anthropic Launches Claude 4.5

PLUS: OpenAI Adds Parental Controls

Author

The Summary AI
September 30, 2025

In partnership with

Welcome back!

Anthropic just dropped Claude Sonnet 4.5, a release built to dominate coding, reasoning, and long-running workflows. With 30+ hour runtimes, strong math scores, and a new Agent SDK, it shows Anthropic wants to be the developer’s first choice for building AI agents. Let’s unpack…

Today’s Summary:

  • 🚀 Anthropic releases Claude Sonnet 4.5

  • 🎶 Suno debuts AI music studio

  • 🛡️ OpenAI rolls out parental controls

  • 🍎 Apple unveils Manzano vision AI

  • đź“Š Microsoft adds “vibe working” to Office

  • đź’ˇ DeepSeek launches V3.2-Exp

  • 🛠️ 2 new tools

TOP STORY

Anthropic releases Claude Sonnet 4.5

The Summary: Anthropic has launched Claude Sonnet 4.5, the strongest model for coding, reasoning, and real-world computer use. The update introduces extended runtimes, with the model maintaining focus for more than 30 hours, and brings new tools like the Claude Agent SDK for building custom AI agents. The release also bundles Claude Code 2.0, API memory tools, and a Chrome extension, in a push to dominate the agent and developer market.

Key details:

  • Anthropic says Claude Sonnet 4.5 built a Slack clone with 11,000 lines of code during a single 30-hour run

  • On SWE-bench Verified, Sonnet 4.5 hit 77.2, up from 72.7 with Sonnet 4, while also earning a perfect score on the AIME math benchmark

  • Early testers reported using it for complex multi-file engineering

  • Experts in finance, law, medicine, and STEM reported Sonnet 4.5 delivered much stronger reasoning than earlier models

  • API pricing remains at $3/$15 per million tokens, more expensive than GPT-5 Codex ($1.25/$10) but cheaper than Claude Opus 4 ($15/$75)

  • Developers get access to a native VS Code extension, checkpointing in Claude Code 2.0, and the Claude Agent SDK to build AI agents

Why it matters: Anthropic is doubling down on coding after OpenAI Codex looked like it was stealing some momentum. The 30-hour runtime shows the model can stay focused to finish complex jobs without breaking mid-task. Anthropic wants to keep developer trust and make its model the best choice for building AI agents.

FROM OUR PARTNERS

A Smarter Way to Write: Just Start Talking

Smart dictation that understands you

Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.

It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.

With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.

Your voice is your strength. Typeless turns it into a superpower.

MUSIC AI

Suno launches AI music workstation Suno Studio

The Summary: Suno has released Suno Studio, a generative audio workstation that blends traditional DAW features with AI-powered stem generation. Alongside it, Suno launched model v5 for Pro and Premier, delivering cleaner mixes, realistic vocals, and higher benchmark scores.

Key details:

  • Suno Studio is positioned as the first generative DAW letting users generate stems inside a multitrack timeline

  • Imports any audio input such as a mix of traditional samples, recorded sounds, and generated tracks, and lets users build new tracks from them

  • Exports both audio and MIDI, for rework in traditional music tools

  • Suno v5 can create complex arrangements with user guidance

Why it matters: Generative music is moving from small experiments into tools that could belong in a professional kit. Suno Studio places AI directly inside the producer’s workflow, as a natural extension of the editing timeline. The consequence is less labor on execution, and more focus on deciding which ideas and directions are worth pursuing.

FROM OUR PARTNERS

How Top Teams Read Between the Lines

Turn customer feedback into evidence that moves your product roadmap faster

Enterpret unifies tickets, reviews, and surveys in one place; auto-tags themes using AI. Helping PMs size impact by revenue; quantify affected users and why; package insights with verbatim quotes stakeholders remember; and track sentiment after launches.

Used by Canva and Notion to align stakeholders and ship higher-impact work, faster.

OPENAI

OpenAI adds parental controls to ChatGPT

The Summary: OpenAI has rolled out parental controls for ChatGPT, giving families an optional way to manage how teens use the AI system. Parents can link accounts, set quiet hours and enable specific features like voice or image. Teen accounts will also carry built-in protections with an alert system notifying parents if ChatGPT detects signs of self-harm in teen usage.

Key details:

  • Default filters block sexual roleplay, extreme beauty ideals, and risky viral challenges

  • A trained review team evaluates flagged conversations before alerts are sent to parents

  • Linked accounts notify parents if a teen unlinks

  • OpenAI is building an age prediction system to apply teen settings automatically

Why it matters: The new parental controls are OpenAI’s attempt to balance user freedom with real-world harm mitigation. The feature lets parents set boundaries while giving teens safe room to explore. The goal is to normalize AI as a tool that supports tutoring, learning, and creativity in households.

TOOLS

🥇 New tools

That’s all for today!

If you liked the newsletter, share it with your friends and colleagues by sending them this link: https://thesummary.ai/